PALINCODE: Recording cell lineage with ternary palindromic CRISPR bits

The paper introduces PALINCODE, a novel CRISPR-based lineage recording system that utilizes efficient ternary palindromic editing to achieve high-density, stochastic state changes in the genome, enabling the reconstruction of deep lineage trees and clonal evolution in both cell culture and in vivo tumor models.

The Gist

Imagine you are trying to reconstruct the family tree of a massive, chaotic family reunion that happened inside a tumor. You have thousands of relatives (cells), but they all look exactly the same, and no one remembers who their parents were or which cousins they are related to.

This is the problem scientists face when trying to understand how cancers grow or how embryos develop. They need a way to write a "family history" directly into the DNA of every cell as it divides, so they can read it later.

Enter PALINCODE. Think of it as a high-tech, biological "ink pen" that writes a secret code into the genome of a cell every time it splits.

The Problem with Old Ink Pens

Previous methods for tracking cell families were like using a pen that could only write "Yes" or "No" (binary: 0 or 1). Or, they were like a pen that would sometimes break the paper entirely (causing dangerous DNA breaks). If you only have two options, you run out of unique combinations very quickly, and the family tree gets blurry.

The PALINCODE Solution: The Three-Color Switch

PALINCODE is clever because it turns a single DNA spot into a three-way switch (a "ternary" bit, or "trit").

Imagine a light switch in your house.

1. Off (0): The light is off. This is the original, unedited DNA.
2. Left (1): You flip the switch to the left. The light turns on a specific color.
3. Right (2): You flip the switch to the right. The light turns on a different color.

How does it work?
The scientists designed a special DNA target that looks like a mirror image (a palindrome). They use a molecular tool (a base editor) that acts like a tiny scissors and glue gun.

• When the tool hits the target, it can edit the DNA on the top strand OR the bottom strand, but usually not both at the same time.
• Once it edits one side, that edit changes the shape of the DNA just enough to "lock" the other side, preventing it from being edited again.

This creates a permanent, irreversible record: "I was edited on the left" or "I was edited on the right." Because there are three states instead of two, the amount of information you can store is massive. It's like upgrading from a binary code (0s and 1s) to a code that uses 0s, 1s, and 2s. This allows them to track much deeper family trees.

The "Hairpin" Hurdle

There was a catch. Because these DNA targets are mirror images, they naturally want to fold over and stick to themselves, like a hairpin or a paperclip. This "hairpin" shape makes it hard for the molecular scissors to find the target.

The scientists solved this by shortening the scissors. Just like a shorter key might fit better in a jammed lock, they found that using shorter guide sequences (18 letters long instead of 20) allowed the tool to cut through the folded DNA efficiently.

Putting It to the Test

The team tested this system in two ways:

1. In a Petri Dish (The 293T Cells):
   They grew human cells in a lab and watched them divide. By reading the "three-way switches" in the DNA, they could reconstruct a family tree that went back 32 generations. That's like tracing a family tree back to your great-great-great-great-great-great-great-grandparents!

2. In a Living Mouse (The Melanoma Model):
   They injected cancer cells (melanoma) into mice. As the tumor grew, the cells kept writing their family history. Later, they took the tumor out, read the DNA, and also looked at what genes the cells were turning on.

   The Result: They didn't just see who was related to whom; they saw why some cancer cells were winning. They found two main "clans" of cancer cells:

   • Clan A: These cells were aggressive and growing fast.
   • Clan B: These cells were good at invading other tissues (metastasis).

   By combining the family tree with the gene activity, they could see exactly how the tumor evolved and which "family members" were the most dangerous.

Why This Matters

PALINCODE is like upgrading from a black-and-white sketch to a high-definition, 3D movie of a tumor's life story.

• It's compact: It packs a lot of information into a tiny space.
• It's safe: It doesn't break the DNA, it just subtly changes the letters.
• It's deep: It can track cell divisions much further back than before.

In short, PALINCODE gives scientists a superpower: the ability to look at a single cell in a tumor and instantly know its entire ancestry and what it's planning to do next. This could be a game-changer for understanding how cancer spreads and how to stop it.

Technical Summary

1. Problem Statement

Reconstructing complete and accurate cell lineage trees is a fundamental challenge in developmental biology and cancer research. While CRISPR-based lineage tracing has advanced, existing systems face significant limitations:

• Information Density: Many systems rely on double-stranded breaks (DSBs) which are cytotoxic and often result in binary (indel/no-indel) or low-diversity outcomes, limiting the amount of information stored per genomic locus.
• Complexity: Prime editing systems offer high diversity but require complex pegRNA design and often suffer from variable efficiency.
• Stochasticity vs. Determinism: Effective lineage tracing requires stochastic editing events to resolve branching trees. However, achieving mutually exclusive stochastic outcomes from a single locus without DSBs has been difficult.
• Scalability: Current methods often require large genomic footprints or multiple guide RNAs to achieve sufficient resolution, making them difficult to scale for deep lineage reconstruction.

2. Methodology: The PALINCODE System

The authors developed PALINCODE (Palindromic Coding and Decoding), a CRISPR-based lineage recording system that utilizes ternary CRISPR bits (cBits).

• Core Mechanism:

  • Palindromic Targets: The system employs palindromic CRISPR target sites where the core sequence is mirrored on both DNA strands, flanked by Protospacer Adjacent Motifs (PAMs) at both ends.
  • Ternary States (Trits): A single cBit can exist in three mutually exclusive states:
    1. State '0' (Wild-type): No editing.
    2. State '1' (Left-edited): Base editing occurs on the forward strand.
    3. State '2' (Right-edited): Base editing occurs on the reverse strand.
  • Mutual Exclusivity: The design ensures that once a base is edited on one strand (e.g., converting an Adenine to Guanine), the resulting Guanine sits in the "seed" region of the opposite strand. This prevents the Cas9 enzyme from binding and editing the opposite strand, effectively locking the state.
  • Base Editing: The system utilizes Adenine Base Editors (ABE), specifically ABEMAX and ABE8e, to perform DSB-free editing, converting Adenine to Guanine.

• Optimization Strategies:

  • Truncated Guides: The authors discovered that standard 20-nucleotide guides formed inhibitory hairpin structures in palindromic sequences. They optimized the system using 18-nucleotide truncated guides, which reduced folding energy and significantly increased editing efficiency while maintaining mutual exclusivity.
  • Sequence Tuning: They identified that the nucleotide immediately flanking the PAM (specifically a Guanine at the +1 position) influences the bias toward left or right editing, allowing for the tuning of editing balance.

• Experimental Models:

  • In Vitro: Human 293T cells were engineered with integrated PALINCODE recorders (PalT7 and PalRNF2 targets) and subjected to single-cell sequencing to validate lineage reconstruction depth.
  • In Vivo: A375 human melanoma cells (harboring BRAF V600E) were engineered with PALINCODE barcodes, transplanted into immunocompromised mice, and tumors were analyzed via single-cell RNA sequencing (scRNA-seq) to jointly read out lineage history and gene expression.

3. Key Contributions

• Novel Encoding Strategy: Introduction of a ternary (trit) encoding system using palindromic DNA targets, increasing information density by approximately an order of magnitude compared to binary indel-based systems.
• DSB-Free High-Efficiency Editing: Demonstration that truncated guides (18bp) can overcome secondary structure issues in palindromic sequences, enabling high-efficiency, mutually exclusive base editing without cytotoxic DSBs.
• Theoretical Capacity: The system achieves a theoretical coding capacity of up to $10^{25}$ bits, enabling the reconstruction of lineage trees 32 cell divisions deep with high accuracy.
• Joint Lineage-Transcriptome Profiling: Successful application in an in vivo melanoma model to simultaneously reconstruct clonal evolution and map transcriptional states, identifying drivers of tumor heterogeneity.

4. Key Results

• Simulation Accuracy: Simulations using quartet distance metrics showed that with 30–50 well-balanced cBit recorders, lineage trees could be reconstructed with <5% divergence from the ground truth.
• Editing Efficiency:
  • Standard 20bp guides showed poor efficiency in palindromic targets due to hairpin formation.
  • 18bp truncated guides achieved >30% editing efficiency for PalT7 and maintained mutually exclusive left/right outcomes.
  • Pre-edited targets (simulating a locked state) showed an order of magnitude reduction in further editing, confirming the "locking" mechanism.
• Single-Cell Lineage Reconstruction (293T):
  • In 293T cells, the system generated lineage trees with a median depth of 32 cell divisions and a maximum of 51.
  • Bootstrapping analysis showed strong branch support (median 87%).
  • Lineage capture rates were high, with a median integration capture rate of 78% across 7,033 single cells.
• In Vivo Melanoma Evolution:
  • In A375 xenografts, the system identified 11 distinct clonal populations from a diverse starting pool.
  • Two dominant clones (Clone 1 and Clone 2) were distinguished by unique transcriptional signatures:
    • Clone 1: Enriched for immune response genes (ISG15, IFI6) and proliferation markers (Ki-67), expressing SPANXB1/SPANXC.
    • Clone 2: Enriched for invasion and metastasis genes (LGALS3, FOXC2, CAV1).
  • Heritability Analysis: The study linked specific gene expression patterns (e.g., TNFAIP8 in Clone 1, LGALS3 in Clone 2) directly to the reconstructed lineage tree structure, revealing how specific clones evolved distinct phenotypes.

5. Significance and Impact

• High-Resolution Lineage Tracing: PALINCODE offers a scalable, high-density platform for mapping developmental and evolutionary histories, surpassing the resolution of many competing CRISPR and recombinase-based approaches.
• Biological Tractability: By avoiding DSBs, the system minimizes cellular toxicity and genomic instability, making it suitable for long-term in vivo studies and therapeutic applications.
• Cancer Research: The ability to jointly profile lineage and transcriptomics in tumors provides a powerful tool for understanding clonal selection, metastasis, and drug resistance mechanisms in real-time.
• Future Potential: The compact nature of cBits suggests compatibility with spatial sequencing technologies. Future iterations could further optimize the left-right balance and integrate with prime editing to increase information content per site.

In summary, PALINCODE represents a significant leap forward in lineage tracing technology, providing a robust, high-density, and biologically compatible method to reconstruct complex cellular histories with unprecedented resolution.